LBBAP Capture Criteria

Definition

LBBAP capture criteria are the ECG, electrogram and pacing-manoeuvre tests used to decide whether a septal lead captures the left bundle branch (LBBP), the adjacent septal myocardium only (LVSP), or a fascicle (LFP). (sources/melos-lbbap-ehj-2022, rating: high)

The distinction matters because the three states can look almost identical on a 12-lead ECG, and because most registries count "LBBAP success" as either LBBP or LVSP. In MELOS, 21.5% of 2,533 implants were LVSP and only 9% were proximal LBBP. (sources/melos-lbbap-ehj-2022, rating: high)

Key Concepts

The capture spectrum

Three sub-types are defined by what is captured and separated by the potential-to-QRS interval and frontal axis. (sources/melos-lbbap-ehj-2022, rating: high)

LFP sub-types are named by frontal axis: LAFP with QRS positive in II and III; LSFP with II positive/isoelectric and III isoelectric/negative; LPFP with QRS negative in II and III. (sources/melos-lbbap-ehj-2022, rating: high)

The operational binary is selective versus nonselective capture. Nonselective LBBP captures the LBB and adjacent septal myocardium; selective LBBP captures the LBB alone. (sources/csp-ehra-2023, rating: high)

The reference standard — direct electrophysiological proof

Every surface criterion is an ECG surrogate: it infers capture from how the paced beat looks. One prospective study (n=30) sidesteps that with three catheters — a temporary His lead, the LBBP lead, and a duodecapolar catheter across the aortic valve — and defines direct LBB capture as a retrograde His potential and/or an anterograde left conduction-system potential (PLCS) during LBB pacing, with Stim-His <35 ms. (sources/lbbp-vs-lvsp-criteria-jaccep-2021, rating: medium)

The internal check is what makes it convincing: Stim-PHis was 21.8 ± 5.9 ms, identical to each patient's intrinsic His-to-LBB-potential interval (21.8 ± 6.9 ms), and it stayed constant across selective, nonselective and high-output pacing (21.4 / 21.8 / 21.2 ms; P = .759). Once the LBB is captured, Purkinje intervals are unaffected by whether septal myocardium is also captured. (sources/lbbp-vs-lvsp-criteria-jaccep-2021, rating: medium)

Graded against that standard:

(All rows: sources/lbbp-vs-lvsp-criteria-jaccep-2021, rating: medium)

Stim-LVAT cut-offs were 75 ms in non-LBBB (SN 82%, SP 95%) and 85 ms in LBBB (SN 76%, SP 93%). The absolute value is not transferable between patients, but within a patient the post-capture value is constant regardless of output. In 53.3%, the paced Stim-LVAT exceeded the native PLBB-LVAT by ≥5 ms, so a longer paced value does not exclude capture. (sources/lbbp-vs-lvsp-criteria-jaccep-2021, rating: medium)

Caveat: the technique "is complicated and may not be practical for routine clinical use", the population was "small to draw robust conclusions", and every 100%-specificity claim rests on 30 patients without confidence intervals. (sources/lbbp-vs-lvsp-criteria-jaccep-2021, rating: medium)

Master comparison

Specificity is quoted for the stated cut-off; where two cut-offs are listed, both values are given.

Output-dependent QRS transition

The practical gold standard is an output-dependent transition in paced QRS morphology: ns-LBBP → s-LBBP, or ns-LBBP → LVSP. (sources/csp-ehra-2023, rating: high)

• ns-LBBP → s-LBBP: V1 RWPT prolongs >10 ms, and EGM splitting appears.
• ns-LBBP → LVSP: V6 RWPT prolongs ≥15 ms — 100% specific for loss of LBB capture.
• Availability is the problem: only 26.4% demonstrated a transition in MELOS, rising to 75.4% at more proximal sites at acute implant. (sources/melos-lbbap-ehj-2022, rating: high)
• Programmed stimulation (S2/S3) is the fallback when the threshold test is inconclusive.
• Continuous monitoring can define the transition more strictly: an abrupt ≥10 ms fall in Stim-V6RWPT over two adjacent complexes at constant output during screwing, or a fixed Stim-V6RWPT with a new isoelectric interval and discrete EGM component. (sources/coi-lbbap-hrs-2024, rating: medium)
• Real-time signs during rotation: sudden V6RWPT shortening ≥10 ms, R′ amplitude increase in V1, new S waves in I/V5/V6, and repolarisation normalising in V5–V6. (sources/csp-jaccep-2023, rating: very high)

Interval criteria

V6RWPT

Absolute stimulus-to-V6-R-wave-peak time is the most widely used single measure. (sources/csp-ehra-2023, rating: high)

• <74 ms is 100% specific in narrow QRS or isolated RBBB; <80 ms is 100% specific in LBBB, IVCD or wide escape rhythm.
• Measured from the pacing stimulus, add ~30 ms latency for the pseudo-delta wave in nonselective LBBP.
• A validated operating point in 124 patients was 83 ms (SN 84.7%, SP 96.3%), with <74 ms conferring 100% specificity. (sources/lbb-capture-criteria-hrs-2021, rating: high)
• Direction matters: the criterion is built on the patient's native V6RWPT. If paced V6RWPT becomes longer than native, LBB capture has not been achieved.

V6–V1 interpeak interval

Using the patient's own RV activation as a reference survives the conduction disease and chamber enlargement that inflate absolute V6RWPT. (sources/csp-ehra-2023, rating: high)

• >44 ms is 100% specific; >33 ms indicates likely capture (SN 71.8%, SP 90.0%).
• The interval is longest in selective LBBP (62.3 ± 21.4 ms), intermediate in nonselective LBBP (41.3 ± 14.0 ms) and shortest in LVSP (26.5 ± 8.6 ms). (sources/v6-v1-interpeak-europace-2022, rating: high)
• It rescued 44/69 (63.8%) cases falsely negative by V6-RWPT >75 ms; the combined rule (V6-RWPT <75 or V6–V1 ≥33) gave SN 78.2%, SP 94.2%.
• It is independent of native QRS type (P = .47) and of pacing latency, because both absolute measures share the same delay.
• It fails in very proximal leads: 14/69 (20.3%) had no V1 RWPT prolongation on transition to selective capture; 4/14 had an LBB potential–QRS ≥30 ms, implying retrograde RBB activation masked the loss of septal capture.
• External validation reproduced the cut-offs: V6–V1 ≥33.5 ms gave SN 77.1%, SP 84.6%, and >44 ms was again 100% specific. (sources/ecg-score-lbb-capture-europace-2023, rating: high)

Physiology-based comparison with the native QRS

Because LBB pacing depolarises the LV physiologically, the patient's own native QRS is the reference for their own paced QRS. (sources/lbb-capture-criteria-hrs-2021, rating: high)

• Two equalities were tested and both held (124 patients, 357 ECGs):
  1. Paced V6 RWPT measured from QRS onset = native V6 RWPT.
  2. Paced V6 RWPT measured from the stimulus = LBB potential-to-V6-R-wave-peak interval.
• Criteria built on these rules achieved sensitivity 88.2–98.0% and specificity 85.7–95.4%.
• The first criterion diagnosed nonselective LBB capture in 98 of 100 patients and misdiagnosed it in 3 of 35 with LVSP.
• It fails where measurement fails: a determinable QRS onset is required, rS morphology in V6 shortens V6RWPT by cancellation, and the reference standard is itself ECG-based.

Combined LBBP score

The individual criteria are specific but insensitive; combining three into a point score recovers sensitivity. (sources/ecg-score-lbb-capture-europace-2023, rating: high)

• aVL-RWPT <79 ms separated capture from LVSP with SN 71.2%, SP 88.4% (<74 ms 100% specific), adding information when the left posterior fascicle is captured and V6 shows an rS pattern.
• The score (0–8): 0/1/2 for V6-RWPT (>83 / 80–83 / <80 ms), 0/1/2 for aVL-RWPT (>79 / 74–79 / <74 ms), and 0/2/4 for V6–V1 (<33 / 33–44 / >44 ms).
• ≥3 points: SN 89.2%, SP 100%, AUC 0.976, reclassifying 10/14 cases the two-criterion rule got wrong.
• Caveat: only the two pre-existing cut-offs were externally validated. The score itself was derived and tested in the same 71 patients, without confidence intervals or outcome data.

Morphology-only

This asks whether the paced QRS shape alone can settle capture, with no measurement at all (n=115; two blinded observers shown only lead V6). (sources/ecg-lbbap-criteria-jaccep-2025, rating: medium)

• V6 upstroke/downstroke pattern: fast/slow (or fast/fast) indicates LBB capture; slow/fast or slow/slow indicates LVSP.

• Ratio <1 behaves almost as a clean binary — only 6 patients had a ratio >1 during proven LBB capture and 4 a ratio <1 during LVSP — giving SN 0.96, SP 0.95, PPV 0.95, NPV 0.96; AUC 0.985. The optimal cut-off was <0.87 (SN 0.91, SP 0.99).
• A second sign — new downstroke slurring/notching in any lead not present during penetration, or an "M" pattern in the inferior leads — was present in 66.1% of LBB capture vs 12.2% of LVSP. Restricted to an inferior paced axis it reached 91.1% vs 13.3% (SN 0.91, SP 0.87), but overall sensitivity was only 0.59.
• It is not V6RWPT restated: the upstroke is measured from the real QRS onset, not the pacing spike, so it excludes the isoelectric interval.
• Blinded validation: 86.7% of LBB captures and 88.8% of LVSP correctly classified with no measurements, interobserver kappa 0.70.
• Where it breaks down: accuracy falls to 70.4% in dilated cardiomyopathy vs 93% non-DCM (P = .004); 5/29 DCM patients (17.2%) showed an LVSP-type pattern during proven LBB capture; in wide QRS the criteria were not superior to V6RWPT <85 or <100 ms.
• Read the cohort first: inclusion required a documented QRS transition and all 115 had one, yet the criteria are proposed "in the absence of QRS transition".

The LBB potential (an adjunct, not proof)

An LBB potential shows the lead is anatomically adjacent to the LBB; it does not prove capture against direct electrophysiological evidence. (sources/lbbp-vs-lvsp-criteria-jaccep-2021, rating: medium)

• Its recorded rate is not a stable property: 14% in the multicentre SDL registry, 40.6% in the programmed-stimulation cohort, 50–80% in the founding review, 65% in the morphology study, and 98.3% in one narrow-QRS cohort. (sources/stylet-lbbap-jce-2022, rating: medium) (sources/lbbap-jacc-2019, rating: high)
• Substrate matters: 89% in baseline narrow QRS versus 33% in LBBB (P = 0.001) within one cohort and one recording setup. (sources/ecg-lbbap-criteria-jaccep-2025, rating: medium)
• Recording configuration matters: unipolar recording during implantation obscures local EGMs, and an LBB potential concealed in LV activation in LBBB becomes visible only when conduction is restored by His corrective pacing. (sources/lbbap-implant-jce-2019, rating: medium)
• The potential is dynamic: it appears gradually and grows in amplitude as the tip advances. (sources/lbbap-implant-jce-2019, rating: medium)

Programmed deep septal stimulation

Deep septal stimulation with LBB capture is almost always accompanied by local myocardial capture, producing a fused nonselective QRS. The insight is refractoriness: the two tissues have different effective refractory periods, so an extrastimulus can encroach on one before the other. (sources/programmed-septal-stim-lbbp-jce-2020, rating: medium)

• The manoeuvre: programmed extrastimuli from the deep septal lead; the sudden change in extrastimulus QRS morphology as each ERP is encroached categorises the response as myocardial, selective LBB or nondiagnostic.
• Yield: 269 stimulations in 143 patients; diagnostic in 114 (79.7%), visualising both components of the paced QRS. LBB potential was visible in only 40.6%.
• The ceiling: 20.3% were nondiagnostic, with no gold standard to adjudicate them.
• Specificity is 100% when a diagnostic response is obtained: both the "myocardial" and "selective LBB" responses confirm that the drive train captured the LBB. The 20.3% nondiagnostic rate limits sensitivity, not specificity. (sources/programmed-septal-stim-lbbp-jce-2020, rating: medium)
• Caveats: single-centre; assessors not blinded; no control group; requires an EP setup.

Was the ceiling the method, or the protocol?

The same author's single-case follow-up argues the ~20% nondiagnostic rate belongs to the 600 ms basic drive train, not to programmed stimulation as such. (sources/denker-pes-lbbap-jce-2020, rating: low)

• The case: an 18-year-old with congenital complete AV block and 14 years of epicardial RV pacing developed pacing-induced cardiomyopathy (LVEF 44%); on output reduction there was no QRS morphology change at all, only loss of capture at 0.5 V.
• Two opposite answers at identical settings: same output, lead position and 280 ms coupling interval, the extrastimulus QRS differed completely depending only on the preceding cycle-length sequence.
• The mechanism is Denker's dissociation (1983): myocardial ERP reflects the cumulative effect of several preceding cycle lengths, whereas conduction-system refractoriness varies inversely with the preceding cycle length. With the immediately preceding cycle length held identical, abrupt shortening lowered HPS RRP 350 ± 29 → 344 ± 29 ms while ventricular-muscle ERP rose 225 ± 21 → 233 ± 20 ms; abrupt lengthening reversed both (HPS 335 ± 30 → 351 ± 35 ms; muscle 223 ± 23 → 213 ± 22 ms; divergence P < .005). (sources/denker-hps-refractoriness-circ-1983, rating: high)
• Short-long sequence → LBB ERP prolonged, myocardial ERP shortened → myocardial-only capture at S3, marked by loss of r′ in V1 and a retrograde LBB potential after the paced QRS.
• Long-short sequence → myocardial ERP prolonged, LBB ERP shortened → selective LBB capture at S3, marked by V1 R-wave augmentation, absence of retrograde LBB potential, a latency interval and a discrete EGM deflection.

• Supporting figure: with a 600 ms drive train, a selective LBB response occurred in only 2 of 143 cases; long-short-short protocols or double extrastimuli during intrinsic rhythm give a selective response "very often" (unquantified).
• What it does not establish: the 150 ms and 40 ms values come from one patient; an ERP gap <30 ms gives an uninterpretable response and can produce a false positive; intra-LBB conduction delay can shift the QRS to a myocardial morphology despite capture.
• Three qualifications from the primary source: the mechanism is demonstrated, not derived; it contradicts the canine literature (Han and Moe found parallel behaviour), reconciled only by arguing the studies measured different properties — excitability versus conduction delay and block; and the cohort had normal conduction with bundle branch block excluded.

The 2019 composite rule and its assumptions

The first widely adopted clinical rule was a checklist rather than a measurement. (sources/lbbap-beginners-guide-hrs-2019, rating: high)

• The five signs: (1) paced RBBB pattern; (2) an LBB potential with a potential-to-ventricle interval of 20–30 ms in non-LBBB intrinsic rhythm; (3) Stim-LVAT (stimulus to peak R in V4–V6) that shortens abruptly with increasing output or is shortest and constant at low and high output; (4) the selective-versus-nonselective EGM distinction; (5) direct evidence from an additional catheter.
• The composite rule: criteria 1 and 2 plus at least one of the latter three.
• Where it fails: both mandatory criteria are weak in the populations that matter. A paced RBBB pattern occurs in 23.4% (non-LBBB) and 44.4% (LBBB) of LVSP without capture, and an LBB potential was recorded in only 14% of a multicentre stylet-lead registry. (sources/lbbp-vs-lvsp-criteria-jaccep-2021, rating: medium) (sources/stylet-lbbap-jce-2022, rating: medium)
• What survives is the procedural logic: match each electrical sign to the step it informs — morphology for site, potential for anatomy, Stim-LVAT for depth, output transition for capture.

What the criteria actually prove

• Sensitivity is a property of the cohort, not of the criterion. Re-tested in one 115-patient cohort with proven capture, the established criteria were specific but insensitive: V6–V1 >33 ms gave SN 0.23, SP 1.00 (against 71.8%/90.0% in the original), V6–V1 >44 ms SN 0.21, SP 1.00, and V6RWPT <75 ms SN 0.53, SP 0.94. (sources/ecg-lbbap-criteria-jaccep-2025, rating: medium)
• The apparent weight behind V6/V6–V1 is one dataset reported twice. The V6–V1 paper and the physiology-based criteria paper both screened 468 patients and included the same 124; the 2022 systematic review counts the cohort twice. (sources/v6-v1-interpeak-europace-2022, rating: high) (sources/lbb-capture-criteria-hrs-2021, rating: high)
• No criterion is validated in the population that lacks the gold standard. QRS transition is available in ~26%; the morphology criteria were derived in patients who all had a transition; the combined score was developed and tested in the same 71 patients. No source supplies a decision rule for the majority in whom neither a transition nor a within-patient comparator is available. (sources/lbbp-vs-lvsp-criteria-fcm-2022, rating: high)
• QRS duration cannot rank capture states. nsLBBP preserves LV lateral-wall activation but produces greater interventricular dyssynchrony than LVSP, while selective LBBP has the longest paced QRS. The surface QRS measures the fusion of LV and RV activation, not the quality of conduction-system capture. (sources/lvsp-vs-lbbp-dyssynchrony-hrs-2021, rating: high)

Contradictions / Open Questions

• The published sensitivity of every criterion depends on the cohort in which it was tested. V6–V1 >33 ms is 71.8% sensitive in its validation cohort and 23% sensitive when re-tested in 115 patients with proven capture; V6RWPT <75 ms falls from 100% specific to 53% sensitive. Both sets can be internally correct because the reference standards and case mixes differ — but a sensitivity quoted without its reference standard is not transportable. (sources/ecg-lbbap-criteria-jaccep-2025, rating: medium) (sources/lbb-capture-criteria-hrs-2021, rating: high)
• The new criteria are derived in the patients who least need them. The morphology criteria are offered "in the absence of QRS transition", but inclusion required a transition and all 115 patients had one. They are derived in the ~26% with a transition and recommended for the ~74% without one; performance falls to 70.4% in DCM and is not superior to existing criteria in wide QRS. (sources/ecg-lbbap-criteria-jaccep-2025, rating: medium)
• Two of the most-cited criteria papers are one cohort. The V6–V1 and physiology-based papers both screened 468 and included the same 124 patients at the same two centres. Their agreement is internal consistency, not replication, and the systematic review counts the cohort twice. (sources/v6-v1-interpeak-europace-2022, rating: high) (sources/lbb-capture-criteria-hrs-2021, rating: high)
• The best-performing combined score has never been externally validated. The 0–8 LBBP score reached SN 89.2%/SP 100%, but was derived and tested in the same 71 patients; only the underlying V6/V6–V1 cut-offs were externally validated. It has no outcome data. (sources/ecg-score-lbb-capture-europace-2023, rating: high)
• Two widely used signs are necessary-but-insufficient, and one behaves backwards. A paced RBBB pattern occurs in all LBB captures but also in 23.4–44.4% of LVSP. Paced QRS duration is longer during selective LBBP than during nonselective LBBP or LVSP, because septal myocardial capture partially fuses with the delayed RV activation; QRS narrowing therefore proves resynchronisation, not conduction-system capture. (sources/lbbp-vs-lvsp-criteria-jaccep-2021, rating: medium) (sources/continuous-pacing-lvsp-cjc-2023, rating: medium)
• The LBB potential rate does not reconcile across studies — and one paper contradicts itself. Estimates include 14%, 26.7%, 40.6%, 50–80%, 65%, 76.3%, 98.3% and "all patients". Substrate explains part (89% in narrow QRS vs 33% in LBBB), recording configuration another part, but one cohort reports 76.3% in one passage and "all patients" in another without reconciliation. A potential proves adjacency, not capture. (sources/stylet-lbbap-jce-2022, rating: medium) (sources/ecg-lbbap-criteria-jaccep-2025, rating: medium) (sources/lbbap-implant-jce-2019, rating: medium)
• Is selective capture worth pursuing, and is it even stable? Selective capture is 100% specific for LBB capture, but it is inflated by repeated testing (90% at implant vs 39.6–54% earlier), falls from ~75% at implant to 30.9% at follow-up, and routine post-implant outputs capture local myocardium in most patients anyway. The Fuwai guide warns that overemphasis on the potential and selective capture increases perforation risk through overscrewing. (sources/lbbp-vs-lvsp-criteria-jaccep-2021, rating: medium) (sources/longterm-lbbap-circep-2021, rating: high) (sources/lbbap-implant-jce-2019, rating: medium)
• The programmed-stimulation nondiagnostic rate is probably protocol-specific, but only demonstrated in one case. The 143-patient study found 20.3% nondiagnostic with a 600 ms drive train. The same author's single-case work shows the LBB-versus-myocardium ERP gap is 55 ms at 600 ms drive but 150 ms with short-long-short sequences, and reverses with long-short-short. A cohort comparison of protocols is what would settle it. (sources/programmed-septal-stim-lbbp-jce-2020, rating: medium) (sources/denker-pes-lbbap-jce-2020, rating: low)

Connections

• Part of concepts/Left-Bundle-Branch-Area-Pacing — the parent overview and clinical evidence
• Related to concepts/Conduction-System-Pacing — the wider taxonomy
• Related to concepts/Cardiac-Refractoriness — the myocardial-versus-conduction-system ERP differential that makes programmed stimulation work
• Related to concepts/ECG-Conduction-Disturbances — the surface patterns being interpreted
• Related to concepts/Electrical-Resynchronization-Metrics — why QRS duration and activation time can disagree
• Relevant to concepts/LBBB-Criteria — identifying the substrate before implantation

Sources

• sources/csp-ehra-2023
• sources/csp-jaccep-2023
• sources/lbbap-jacc-2019
• sources/lbbap-beginners-guide-hrs-2019
• sources/melos-lbbap-ehj-2022
• sources/lbbp-vs-lvsp-criteria-jaccep-2021
• sources/lbbp-vs-lvsp-criteria-fcm-2022
• sources/v6-v1-interpeak-europace-2022
• sources/lbb-capture-criteria-hrs-2021
• sources/ecg-score-lbb-capture-europace-2023
• sources/ecg-lbbap-criteria-jaccep-2025
• sources/programmed-septal-stim-lbbp-jce-2020
• sources/denker-pes-lbbap-jce-2020
• sources/denker-hps-refractoriness-circ-1983
• sources/continuous-pacing-lvsp-cjc-2023
• sources/lvsp-vs-lbbp-dyssynchrony-hrs-2021
• sources/stylet-lbbap-jce-2022
• sources/lbbap-implant-jce-2019
• sources/coi-lbbap-hrs-2024
• sources/longterm-lbbap-circep-2021